ALF meeting 5/22/98 in Austin

Participants:
	Mike Andrews		Mentor Graphics
	Archie Lachner		Mentor Graphics
	Jay Abraham		SI2
	John Beatty		IBM
	Tim Ehrler		VLSI Technologies
	Kevin Grotjohn		LSI Logic
	Tim Jennings		Motorola
	Yatin Trivedi		Seva (per phone)
	Cary Wei  		Fujitsu
	Toru Toyoda		NEC
	Wolfgang Roethig	NEC

Topics:

* ALF/OLA roadmap

Scheduled deliverables and identified a lot of dependencies.
We agreed on a phased development approach.

	Organization	Deliverable

	Cadworx		ALF to DCL translator
	IBM		DCL compiler with new APIs
	ASIC companies	OLA libraries for production
	ALF work group	ALF 1.1 draft spec.
	OLA work group	OLA 1.0 draft spec.


A.I.	Copy the Gant Chart from Whiteboard into PowerPoint
	for review.
	- Jay Abraham


* review of ALF 1.0.3

BNF has been corrected so that all objects are uniformly terminated
either with ; or { ... }

Arithmetic models are presented in table form with header rows:
keyword / value / unit / explanation

Will stage ALF 1.0.4 by June 8 and have a conference call June 10
to agree on ALF 1.0.4 to be distributed in OVI booth at DAC.

A.I.	stage updated spec. on web by 6/8
	- Yatin

	set up conference call for 6/10
	- Wolfgang

	make examples (chapter 4) conform to updated BNF
	- Yatin, Wolfgang offers help, if needed

	provide Syntax and semantics for ALF_version annotation
	- Tim Ehrler

	All annotations in information container (3.6.1.5)
	should be quoted string
	- Yatin

	explain semantics of 2-dimensional indexed pins
	-Yatin 

	add explanations to all annotations
	- Yatin

	correct 3.6.5.7
	- Wolfgang

	remove 3.6.5.8
	- Yatin

	change temperature model examples (pp. 4-20)
	- Wolfgang

	provide seperate column for "unit" and "base measurement"
	of arithmetic models
	- Yatin

	present annotations other than arithmetic models
	in table form as well
	- Yatin


* Review of ALF SI note

Chapter 1 (ALF1.0 capabilities) should be added to the examples in ALF spec.

Feedback from Mike Andrews: Point out how "current" for the use by power analysis
can be distinguished from "current" for the use by signal integrity tool.

Chapter 2 ( proposal for extensions in ALF 1.1) need to be reviewed
by the signal integrity experts of the ASIC and EDA companies.

A.I.	copy chapter 1 as chapter 4.13 into ALF spec.
	- Yatin

	identify SI experts and give SI doc to them for review
	- each team member


* Semantics of vector expressions for simulation

The concept is explained in chapter 2, but a more rigorous semantics
definition may be needed.

A.I.	read chapter 2 throroughly and point out flaws
	- each team member

	write a draft for rigorous interpretation of vector expression
	by simulator
	- Wolfgang

* issues with core modeling

How can parametrizable blocks and cores be modeled (e.g. N-bit adder
with N as a parameter)?
Today, only indirect modeling is supported: define the core as a template
and then instantiate the template with fixed values for the parameters.
See introduction to chapter 3-10 for the methodology. Same methodology
can be used to describe extensible primitives using ALF language.

Requests for modeling enhancements from Kevin Grotjohn:
- WIRE object may be allowed inside a CELL for the case that
  CELL is a core or block with its own wireload model.

- "number of bits", which is a parameter of a parametrizable core
  may be allowed as argument of an arithmetic model.
  e.g. delay = f(number of bits) etc.

- allow RC-graphs instead of lumped capacitance attached to a pin

A.I.	establish requirements for core modeling, provide examples, if possible 
	- each team member

	look at existing modeling capability of parametrizeable cores
	using templates and make an assesment of its usefulness and
	its limitations
	- each team member

	look at SPF for RC-graph modeling capabilities
	- each team member (volonteers welcome)

	Try to come up with ALF-style syntax for RC graphs
	- each team member (volonteers welcome)


Next meeting will be July 22 in Phoenix, Arizona, after OLA meeting on July 20-21